Verifying stochastic process containing summation of Martingales is a Martingale

80 Views Asked by At

Let $M=(M_t)_{t \geq 0}$ be a Martingale with respect to the filtration $\mathcal{F}=(\mathcal{F}_t)_{t \geq 0}$. Assume that $\mathbb{E}(M_t^2)<\infty$ for all $t \geq 0$. Let $0=t_0<t_1<...<t_N=T$ be a partition of interval $[0,T]$. Further, let $(\xi_n)_{n=0,1,...,N}$ be a family of random variables such that $\xi_0 \in \mathcal{F}_{t_{0}}$ and $\xi_n \in \mathcal{F}_{t_{n-1}}$ for all $n \geq 1$, and such that $\mathbb{E}(\xi_n^2)<\infty$ for all $n \geq 0$.

Consider the process $X=(X_t)_{t \in [0,T]}$ given by:

$$ X_t=\xi_0+\sum_{n=1}^{\{t\}}\xi_n(M_{t_n}-M_{t_{n-1}}) $$

Where $\{t\}$ is defined as $k-1$ if $t \in (t_{k-1},t_k]$

The problem is to prove the above process is a Martingale with respect to $\mathcal{F}$. I have already proven the first condition (measurability), but I am having trouble proving integrability and the Martingale property. Any help would be much appreciated.

1

There are 1 best solutions below

1
On

For the integrability use the triangle and Hölder inequality. It should be straight forward.

For the Martingale property use that for all $t_{n-1}<t$ $\xi_{n}$ is $\mathcal{F}_{t}$ measureable, so by using the tower property of conditional expectations a finite number of times, you can pull it out of the conditional expectation and by the matringale property of $M$ the most inner difference equals zero. That's why all terms with higher index disappear.


Sry, for the short answer so far. If it is not sufficent, I can answer in more detail later.